A dynamic model of wrist-worn rotational electret energy harvester: Theoretical investigation and experimental validations

Wrist-worn energy harvesters have great potential to sustainably power wearable devices for the Internet of Things. Previous research improved their energy conversion efficiency by structural designs and smart materials, however, arm swing motion's influence on their power generation was underexplored. To fill this gap, we establish a dynamic model of the human-wrist-energy harvester during human walking. Our model highlights the positive impact of body accelerations and arm swing motions on a rotational electret energy harvester (REEH). Leveraging a dataset of 300 participants (150 female and 150 male) and diverse machine learning approaches, we establish an accurate data-driven model for parameters prediction of arm swing motions by Random Forest regression (RFR) method. Furthermore, experimental results demonstrate the proposed model outperforms the previous linear regression model for the prediction of arm swing motions, additionally, the output power is 0.46 mW under the normal walking speed (1.45 m/s) which is sufficient to power a wearable device. Moreover, the numerical results reveal that adjusting the surface voltage or gap distance can enhance the output power for the realization of a self-powered wrist-worn device. These findings provide theoretical guidance for future design and optimization of wrist-worn energy harvesters.